🔌
New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
// RUN — FITS
Can the NVIDIA DGX Spark run Phi-4-Multimodal?
Yes — here's how fast and up to what context, with real measured numbers.
Phi-4-Multimodal on the NVIDIA DGX Spark
| Quant | Fits? | Max context | Speed | Beyond context |
|---|---|---|---|---|
| FP16 | yes | 782k | 17.8-21.2 t/s | ~0.7-0.8 (slow) |
| Q8 | yes | 825k | 35.6-42.4 t/s | ~1.3-1.6 (slow) |
| Q4 | yes | 844k | 63.5-75.7 t/s | ~2.3-2.8 (slow) |
Estimate (memory-bound). Beyond "max context" the KV cache spills to SSD swap and speed drops to the "beyond" figure.
FAQ
Can the NVIDIA DGX Spark run Phi-4-Multimodal?▶
Yes. At Q4 it generates ~63.5-75.7 tok/s and fits up to 844k context; beyond that it spills to SSD swap and slows to ~2.3-2.8 tok/s.
How much context fits?▶
At Q4, up to about 844k tokens stay in fast memory; longer context spills to SSD swap and throughput collapses.